1

TWO DOGMATISTSby

Charles R. PigdenUniversity of Otago

Grice and Strawson's 'In Defense of a Dogma is admired even by revisionistQuineans such as Putnam (1962) who should know better. The analytic/syntheticdistinction they defend is distinct from that which Putnam successfullyrehabilitates. Theirs is the post-positivist distinction bounding a grossly enlargedanalytic. It is not, as they claim, the sanctified product of a long philosophictradition, but the cast-off of a defunct philosophy - logical positivism. The fact thatthe distinction can be communally drawn does not show that it is based on a realdifference. Subcategories that can be grouped together by enumeration will do thetrick. Quine's polemical tactic (against which Grice and Strawson protest) ofquestioning the intelligibility of the distinction is indeed objectionable, but hisargument can be revived once it is realized that 'analytic' et al. are theoretic terms,and there is no extant theory to make sense of them. Grice and Strawson'sparadigm of logical impossibility is, in fact, possible. Their attempt to definesynonymy in Quinean terms is a failure, nor can they retain analyticity along withthe Quinean thesis of universal revisability. The dogma, in short, is indefensible.

I. Introduction'Two Dogmas',1 Quine's great attack on the analytic/synthetic distinction, caused quitea stir when it first came out, and provoked a spate of replies. One of the few to havestood the test of time is H. P. Grice and P. F. Strawson's (1958) 'In Defense of a Dogma'(henceforward DD). It is much anthologized, praised by Quinean revisionists such asHilary Putnam (1962) (1975, p. 34) and appealed to by the likes of R. M. Hare, whosesystems might not survive the demise of a sharp distinction (1981, p. 81). If theanalytic/synthetic dogma is still a going concern, that is largely due to Grice andStrawson's defence.

What are its merits? According to Putnam (1975, pp. 34-35), GS (as Grice and Strawsonwill henceforth be known) offer theoretical reasons for supposing that the distinctionmarks a real difference (which is, after all, the point at issue). They don't just beat theirbreasts and declaim that the difference is real and they feel it in their bones - no, theyargue (1) that there must be a difference for philosophers to succeed in making thecommunal distinction (DD, pp. 42-43).

In fact, GS do more than this. They go on (2) to criticize Quine's central argument that

2

the notion analytic cannot be clearly defined, since it forms part of a closed circle ofconcepts, all of which stand in equal need of clarification. Even if this is true, it does notconstitute a decisive argument either against the category or the distinction marking ifoff. The (unspecified) standards of intelligibility to which Quine appeals are set far toohigh. Whatever they are, it is clear that too many words would fail to meet them (DD,pp. 147-9). (3) GS make a rather unfortunate attempt to break into the circle with aconcrete example of a logical impossibility - unfortunate, because, as I shall argue, it isprecisely the kind of analyticity ('it is analytic that not . . .') that experience mightpersuade us to discard. (Their idea seems to be that a stream of such examples, together,perhaps, with the appropriate contrasts, would yield a sort of inductive definition. DD,pp. 150-1). Finally, they turn to the last and best third of Quine's paper, where he linksthe analytic/synthetic distinction with empiricist reductionism (the second of the twodogmas). Because our statements face the bar of experience 'not individually, but onlyas a corporate body' (1961, p. 41), Quine argues that no single statement can be assignedits peculiar verification conditions. Hence synonymy (analyticity and all the rest) cannotbe defined in terms of co-verifiability. And since it is optional which statements werevise in the face of 'recalcitrant' experience, analyticity cannot be defined as absoluteunrevisability. At best, there is a continuum between those statements we would reviseat the drop of an experience, and those we would give up only if things got reallyweird. GS suggest (4) that synonymy can be defined despite the lack of determinateverification conditions. Two statements are synonymous if, on certain assumptions,experiences which would confirm or disconfirm one, would confirm or disconfirm theother to the same degree (DD, p. 156). And (5) the category 'analytic' can be retainedthough no propositions are absolutely unrevisable, so long as a distinction is preservedbetween those statements that are dropped through mere falsity and those which areabandoned because the concepts involved are changed or given up (DD, p. 157).

This paper is an attack on the defence. But it should not therefore be construed as adefence of the attack. I do think there is an analytic/synthetic distinction - or at leastthat such a distinction can be drawn. But the category of the analytic is small - muchsmaller than is commonly supposed. It is scarcely worth a philosopher's seriousattention. Putnam (1962), (1975), the arch-revisionist of Quineanism, got it more or lessright. He defines a narrow and uninteresting realm of analytic truths consisting of

3

synonymies and their consequences. This is not (Putnam himself to the contrary) thebloated empire of the analytic defended by GS. So before turning to the analyticimperialism of GS, I shall outline Putnam's severely reduced realm and explain why theboundaries had to be pushed back.

The result is that the category of the synthetic - usually defined as the non-analytic - willtend to break up. For being non-analytic becomes a trait too common to conferphilosophic distinction. The items that would be included in the enlarged synthetic aretoo disparate for the grouping to be of interest. Moreover, in so far as 'synthetic' didhave any connotation over and above 'non-analytic', it tended to exclude suchpropositions as the truths of logic and mathematics, now in the non-analytic basket. Soeither the term will be dropped, or it will be supplied with a new connotation, whichdetermines a smaller and more cohesive class. Logic, mathematics, and much else willbe left in no-man's land. It transpires that if Putnam is right about the analytic, GSs'first argument is not as good as Putnam himself supposes. And unfortunately for them,he is.

II. The Truth about AnalyticityPutnam defines analytic truths thus:

Analytic truths are (or are like) statements which fulfill (or nearly fulfill) the followingcriteria, or are consequences thereof:

(i) The statement has the form 'Something (someone) is an A if and only if it (he,she) is a B' where 'A' is a single word.

(ii) The statement holds without exception, and provides us with a criterion forsomething's being the sort of thing to which the term 'A' applies.

(iii) The criterion is the only one that is accepted and employed in connectionwith the term.

4

(iv) The term is not a law-cluster term.

(Putnam [1975], p. 65)

This is not the place to rehearse all Putnam's arguments and elucidations. But the classhe is trying to capture consists of such inanities as 'Bachelors are unmarried men' -boring lexicographic synonymies and their consequences; statements which 'cut nophilosophic ice' (1975, p. 36). He wants to isolate these because they are the onlypropositions which satisfy the ideology of analyticity; the purported accounts of what(bloated) analyticity is. They are (a) true in virtue of (something like) linguisticstipulation; and (b) unrevisable given the language - i.e. a change in assigned truth-value betokens a mere or trivial meaning change (that is all that has happened). Putnamaccepts Quine's argument in 'Truth by Convention' that 'the radical thesis that logic istrue by language alone quickly becomes the harmless truism that "logical truth is trueby virtue of language plus logic"' (Putnam [1962] [1979, p. 188], Quine [1936] [1976]).And if logic is not true by convention, neither is mathematics.

It is not just that logical truth transcends convention - its change transcends conventiontoo. Logic (at least some of it) is revisable. Indeed, Putnam wants to revise it. But 'toassimilate the change from one system of logic to another to the change that would bemade if we were to use the noise "bachelor" to stand for "unmarried woman" instead of"unmarried man" is assimilating a mountain to a molehill' (1975,p. 51). For instance(Putnam thinks), we should shift to a quantum logic rather than accept a physicallyweird or weirdly anti-realist physics. The shift allows us to reject these weirdnesses asuntrue and compels us (in the small) to restrict our inferences. It is not as if we arereading '&' for 'v' and vice versa! (1979, pp. 189-92). Finally, Putnam looks askance at agood many 'analyticities' going the rounds. Many are not even true, let alone true in

virtue of lexicographic conventions or synonymies.2, 3

In fact, Putnam's definition exceeds his brief. Besides its intended victims, it entraps thetruths of logic (and much of mathematics as well). Since they (or at least some of them)are necessarily true (true under any appropriate interpretations), then they are true

5

under interpretations which verify 'All bachelors are unwed' (etc.). Hence they areincluded among its consequences. The analytic has reconquered logic! But Putnamneed not despair since the reconquest is very shaky. For one thing, it may be a matter ofdual sovereignty. Any attempt to define the synthetic as the inclusive consequences ofconsistent subsets of a base class of selected statements, will likewise rule in logic.Moreover, the analytic can be thrown back by fiat. We simply restrict the analytic tothose consequences of (i)-(iv) statements that are not logically (arithmetically,mathematically) true. More interestingly, we could restrict the consequence relation tosomething of less than classical strength. Putnam should sympathize with this. He isnot only a revisionist logician, but specifies that the 'if and only if' of (i) (together withany 'ors' that may be involved) are not the truth-functional creations of classical logic.Presumably they would determine a weaker consequence relation; perhaps one thatwould exclude the truths of logic and mathematics.4

I come now to my first polemical point. If Putnam is right about the boundaries of theproperly analytic, then he is wrong to commend GS. His point is that the onlyideologically sound analyticities are the trivial ones he delineates. These fit the analyticbill: others do not. There is a distinction between true analyticities and others but itdoes not lie where it is usually left. The valid distinction is distinct from the one that iscommonly drawn. GSs' point is that since philosophers in fact agree on the currentdistinction, there must be (pending stupendously good counterarguments) a realdifference there - that is, there must be two distinct categories each bound together bysome significant and remarkable feature. We cannot distinguish the (current) analyticfrom the synthetic simply on the basis of enumeration. (The classes are infinite orindefinitely large.) Thus the current categories must be genuine, possessed of a rationale,if there is to be a consensus in sorting.

For Putnam (though he does not clearly realize this) the practice to which GS appeal issuperstitious. What philosophers do (in terms of drawing distinctions) is inconsistentwith the stories they tell about it (e. g. everything on the larger analytic side is 'true bylinguistic convention' or something similar). What philosophers call the 'analytic' is notdistinguished by any philosophically interesting property (unless being called 'analytic'by 'fifties philosophers counts as interesting). Rather, there is a bundle of disparate

6

subcategories, plus or minus various odds and ends, held together by a myth. Muchthe same must be said of the synthetic. It should fall apart. For (except for hard-lineempiricists) what unified this class was the lack of this mythical analyticity - in itselfsupposedly a significant feature. All that remains now is a class some of whosemembers are marked by greater responsiveness to experience, a more clear-cut concernwith 'matters of fact and existence'. Thus the philosophic practice is based on anillusion. T here is no single significant something about the 'fifties analytic whichdistinguishes it from the synthetic (and vice versa).

But according to GS this is precisely what cannot be, given the philosophical consensus.So, if GSs' argument is correct, Putnam is in trouble. He cannot commend theirargument and ignore its conclusion - that there is an analytic/synthetic distinction otherthan the one drawn by him. The loyal Putnamite must therefore improve upon the Masterby showing how this other distinction can be drawn in the absence of a real difference.How did philosophers (at least philosophers of the 'fifties) do the trick? Whence thisconsensus on two unreal categories?

The answer lies shrouded in the mists of time, in a past that by 1958 was alreadyforgotten.

III. History of a DistinctionGS open up their argument (1) with a whopping historical falsehood. The distinctionexpressed by analytic/synthetic was what, in earlier times, philosophers 'havesupposed themselves to be expressing' by such pairs as 'necessary' and 'contingent', 'apriori' and 'a posteriori', 'truth of reason' and 'truth of fact'. The distinction is not just afad of GS & Co., but is grounded in a long and 'not wholly disreputable' tradition (DD,p. 142). This is sheer nonsense. The necessary/contingent distinction goes back toAristotle at least, but only coincides with the analytic/synthetic in the writings ofempiricists and their followers. The analytic and the synthetic were unknown to thescholastics, but if we try, anachronistically, to impose these notions on their work, wecan see that the necessary and the analytic do not come to the same thing. God'sexistence was their paradigm of necessity, but according to Aquinas it was (for us)

7

synthetic.5 (The five ways all work from synthetic premises.)6 The basic principles ofmorality (natural law, etc.) are necessary, grounded in God's immutable will, but hardlyanalytic. Indeed, necessary but non-analytic moral truths persist through Price, Reid,and Kant right up to Ross in 1939!7 Necessities abound in Descartes, Spinoza, andLeibniz, but it would be a bold scholar who maintained that, if properly analyzed, theywould all wind up analytic. And no one would say that that is how Descartes saw them!Quine is notoriously opposed to both the necessary and the analytic, but being opposedby Quine does not convert them into the same thing. Kant, of course, invented theanalytic and the synthetic, as well as the a priori and the a posteriori in their modernmeanings. But again, the distinctions are conceptually and extensionally distinct. Thesynthetic a priori is a capacious realm, including morality, arithmetic, and geometry.

(Its truths are necessary too.) Schopenhauer8 and even Frege9 were Kantians in this,

though Schopenhauer had his doubts about morality,10 and Frege tried to expand theanalytic to include arithmetic. Even so, geometry remained. Russell retained thesynthetic a priori right up until 1912 (after Principia Mathematica) and the old-style

intuitionists never gave it up.11 Even empiricists, anxious to do away with such conceptsas the synthetic a priori, have not all resorted to the analytic. Mill notoriouslyconverted the truths of arithmetic and (some) axioms of geometry into high-level

empirical generalizations.12 In short, the only traditional figure to 'express' GSs'distinction clearly is Hume with his (contingent) 'matters of fact or experience' and his

(necessary) 'relations of ideas' (covering arithmetic and geometry).13 Locke perhapsprefigures him. He has an almost Putnamite category of 'trifling propositions', plus theset of intuitive truths and their (demonstrative) consequences. (This contains arithmeticamong other things.) But since he does not assimilate the one to the other, he lacks the

modern concept of analyticity.14 Leibniz sometimes seems to be running a GS-type line,with his (necessary) truths of reason and (contingent) truths of fact (though hiscognitive account of morals complicates matters). But the weirdness of his metaphysics

led him astray. In the end there are no contingent truths.15

The Vienna Circle revived the philosophy of Hume, and with it his distinction. Quapositivists (descendants of Comte) they were concerned to show that everything worthknowing, all truths about the world, could be known by the scientist. The synthetic a

8

priori was denied, and morality dismissed as non-cognitive, since if there were truths ofeither kind they could presumably be known by non-scientific means. This would open

the way for the free-wheeling metaphysics they so detested.16 Thus the synthetic andthe empirical were identified. Mathematics was a problem, but logicism allowed themto shunt it into the analytic. And by marking out the analytic as the philosopher'sstamping ground, they resolved a conflict within the positivist soul. The logicalpositivists possessed a strong streak of self-abnegation, even self-immolation. Since thescientist could know everything worth knowing, they were inclined (as merephilosophers) to consign themselves to the flames as containing nothing but sophistryand illusion. But few can write off their own life-work as worthless, and still fewercomfortably write themselves out of a job. Thus the positivists devised a field ofinquiry where the results, though incomparably less significant than those of science,were yet not devoid of interest or utility. Analytic truths were (or were reducible to)synonymies or truths of logic, and were hence (in an extended sense) tautologies. Theysaid nothing about the world; thus the philosopher (whose stock-in-trade they were)was not fit to fasten the shoestrings of the scientific Messiah. But in the explication ofconcepts the positivist found a function, especially since what constituted theverification conditions of a proposition was an analytic question and hence the

philosopher's preserve.17 Unworthy as he was, the philosopher could prepare the wayfor the scientific Lord and make his paths straight. With the proud humility of Lockeanunderlabourers, they set to work. The results are too well known. Their greatestachievement was the education of their philosophical executioners.

Philosophers of the post-war world retained the distinction but dropped the theory.Explicit verificationism went out of vogue (though throughout the 'fifties verificationist

arguments abound).18 Thus there was no longer any reason to identify the syntheticand the empirical. But people did not stop. With the abandonment of verificationism, alarge class of analyticities were eliminated - those pairing theoretical statements withtheir verification conditions. Nobody noticed. The original reasons for consigningarithmetic to the analytic no longer held. Godel had shown that no set of axioms could

deliver all the truths of arithmetic. Logicism had required just this.19 Besides, therewere rival ontological bases for arithmetic, alternative set theories positing the existence

9

of different entities. On Kantian criteria, this would tend to make arithmetic synthetic

and not analytic after all. Only Quine seems to have realized this20 - which perhapsexplains his zeal to abolish both categories.

But the official boundaries of the analytic did not shrink. On the contrary, there was acertain amount of analytic creep. A. J. Ayer, finding himself intellectually unfitted toexplicate the concepts of science (the task he had set philosophers in Language, Truth andLogic), was lucky: 'It began to seem to me that there were still problems in the theory ofknowledge that needed further attention: and so I pursued this easier course' (1977, p.164). Most philosophers lack Ayer's charming insouciance. They need an elaborateideology to justify their academic existence. Hence the boom in 'conceptualinvestigations' conceived as an inquiry into the larger analytic. The category had to beexpanded to assure philosophers of an interesting supply of problems - and to validatetheir solutions. The later Wittgenstein was a malign influence here. The general trendof his later philosophy is to blur the analytic/synthetic distinction. For example (PI,I.242), 'If language is to be a means of communication there must be agreement not onlyin definitions but (queer as this may sound) in judgements'. Understanding the worldand understanding language cannot be torn apart. A sufficiently radical change inworld-view would betoken a change in our linguistic capacities - and vice versa (see Pl,II.xil). But his conception of philosophy as sharply distinct from science and operatingat a different level, as concerned with 'grammar' and 'what is possible before all newdiscoveries and inventions' (Pl, I.126) - all this required a sharp analytic/synthetic split;and an enlarged analytic to accommodate the expanded range of questions considered.The self-refuting thesis that we do not (cannot? should not?) advance theses inphilosophy (Pl, I. 128) probably blinded him to the nature of the epistemic backingneeded for the theses he advanced. Unfortunately his followers preferred illusion toinsight. By the time of DD the most amazing claims get labeled 'analytic'. Malcolm, forexample, thinks it conceptually impossible (i.e. analytically impossible) that the worldcould have come into existence five minutes ago complete with memory traces,remains, etc. (1963, pp. 187-202). He is honest enough to admit that his analyticities arenot analytic in the old positivist sense. The five-minute world is not a 'self-contradictory' hypothesis. But 'there is not just one brand of logical impossibility' (1963,

10

p. 201).

It is not just extravagances such as this, but precisely this extravagance (among others)that Putnam's restricted analytic is designed to curb. (See Putnam [1975, p. 37], wherethe unnamed philosopher is either Malcolm or someone very like him). Putnam'sexplicit attack on Malcolm (1962) (1975, p. 306) censures his reliance on 'the idea of adepth grammar which provides a "fact/convention" dichotomy on a level inaccessibleto ordinary lexicographic investigations'. Quine, too, is no friend to an expandedanalytic.

Marx maintained that the evils and abuses of capitalism could only be remedied by itsabolition, while Bernstein proposed that it be controlled, confined, and refined bysocialist politicians. Quine maintains that the evils and abuses of the analytic/syntheticdistinction can only be remedied by its abolition, while Putnam proposes that theanalytic be controlled, confined, and refined by realist philosophers. But whetherrevolution or reform is the right option, there was (in the 'fifties and early 'sixties) acrying need for one or the other. Philosophy was dominated by a dark distinction withnothing more by way of backing than the ghost of a dead theory. In the shadow of thisspecter philosophical superstitions flourished. The distinction was not the sanctifiedproduct of a reputable tradition, but the decayed cast-off of a defunct philosophy. Theestablished use to which GS appeal is a 'fifties artifact - and a rather shoddy one at that.

IV. Distinctions Without a Difference: How to Make ThemBut an artefact, however shoddy, is still real. So the question remains: How didphilosophers draw the distinction without a difference? If there is no significantcharacteristic common to the members of at least one category, how can we have agreedin consigning propositions to one category or the other?

GSs' argument runs as follows:If a finite collection is divided into two classes, the distinction need not mark a realdifference. Being an A rather than a B need amount to nothing more than being a, b, orc as opposed to d, e, or f. Some A's might resemble some B's more than they resemble

11

their classmates. You could still tell A's from B's simply by knowing all the members ofboth groups. The situation is different when the classes are open. If speakers can agree,more or less, in distinguishing A's from B's when both have a potentially infinitemembership, there must be something about A's and B's which enables them to do so.Both the categories and the distinction must have a rationale. The members of at leastone class must share some (more or less) discernible characteristic. It does not evenmatter if there is a disputed area, so long as there is general agreement on which casesare doubtful. Philosophers agree in distinguishing the analytic from the synthetic.Therefore there must be something about the items distinguished that makes thispossible. The classes reflect genuine categories and the distinction a real difference.

This argument is inconclusive. The categories and the distinction might lack a rationaleeven though we can agree on sorting. Maybe the two classes are themselves composedof open subclasses which do have a rationale. These must be finite in number and itmust be easy to tell whether something is a member. Then, because we can determinethe membership of the subclasses, and because the community has been taught byenumeration which are included in A and which in B, we can agree on which items aremembers of A and which of B, even though there is nothing non-arbitrary connectingthe subclasses which comprise them. Being an A rather than a B, is to be a member of a,b, or g, all of which are included in A, rather than a member of d, e, or z, all of which areincluded in B. And to be a member of a, b, or g, is to possess one of the properties, a’, b’,or g’. But for all that, members of a, might resemble members of d, e, or z, more closelythan they do the rest of the A's.

Such was the legacy of positivism in the 'fifties and early 'sixties. Mathematicalpropositions, truths of logic, synonymies and their consequences are all classes whosemembership can be agreed on - especially if no one looks for borderline cases.Commonsense observations ('It is raining'), scientific statements ('This is molybdenum')are likewise easily recognized. Propositions that cannot be fitted into these classes canbe swept under the carpet. Many will be regarded with suspicion anyway, asmetaphysics has not yet recovered from the blows dealt by verificationism (which isstill treated with respect). Philosophers have been taught to include the first group ofproposition-classes in the analytic and the second in the synthetic. But we have no

12

proof that there is anything non-arbitrary binding the two groups together, especially aswe have a historical explanation - the prevalence of positivism - of how the groupingsoriginally arose. A diet of bad examples ('This is molybdenum' rather than 'E = mc2' or'All reptiles are cold-blooded') has obscured the fact that some easily identifiablescientific propositions do not fit the (rather vague and confused) synthetic bill. (Forexample, they are not in any straightforward way refutable by experience, or do not dealwith the properties of existent things.) Sheer ignorance has concealed the fact thatlogicism has failed (or is at least in difficulties) and thus that the original reason forclassing mathematics as analytic no longer holds.

The consensus on which GS rely is, I suggest, much exaggerated. But in so far as it didexist, it was a hang-over from a series of false beliefs. The existence of open subclasseswith discernible rationales allowed it to persist. But the sorting practice was devoid ofphilosophic justification. Once the beliefs which bound the subclasses together havebeen abandoned, the categories - and the practice - should be dropped or reassessed.

V. Unintelligibility and Orwellian PhilosophiesArgument (2) is a better bet. GS are right to query Quine's standards of intelligibility.In effect, Quine demands that a concept X only be accounted intelligible (i.e. allowed tomake sense) if an explanation can be provided which meets the following criteria: (1) Itprovides necessary and sufficient conditions for something's being X (it is an 'if andonly if' definition); and (2) the concepts in the explanandum do not themselves involveX (i.e. X would not occur in their explananda). GS suggest that a good manyindispensable concepts would fail this test. And at present no explanations which meetthe criteria are forthcoming for the concepts 'morally wrong', 'blameworthy', 'false','fact', 'denial', and 'assertion' (DD, pp. 147-8). Quine's demands are tendentiouslyexcessive.

In fact we can go further. Quine's argument here represents Orwellian philosophy at itsworst. The anonymous tyrants of 1984 sought to enforce a 'canonical language',Newspeak, in which subversive thoughts could not be expressed. Empirical andanalytic philosophers since Hobbes have sought to eliminate philosophical opposition

13

wholesale by enforcing a canonical language (usually a restricted version of English) inwhich subversive thoughts (scholastic, rationalistic, or metaphysical thoughts) cannotbe expressed. Without state-power this looks like a pathetic fantasy, but in fact it hasbeen a remarkable success, usually because Orwellian philosophers have marched intime with the progressive ideologies of the age. The original empiricists (Hobbes,Locke, Berkeley, and Hume) grounded language in a psychology of ideas conceived ascopies of sensa. Any word to which no such idea corresponded was meaningless. ThusHume, for example, disposes of necessary causal connections. The weakness of thisexplicit approach is that the Orwellian theory is exposed to criticism. Indeed, theconcepts it is designed to dispose of can be revived as counterexamples. Price (1758)(1974, p. 25), points out that we do have the notion of a necessary connection, and thatthe psycho-linguistic theory which rules this out is therefore false. The Orwellianprogram of the Vienna Circle suffered much the same fate. The linguistic theory wasexplicit, and as such subject to a damning critique. Post-war Orwellians are wiser. Thetheory of meaning, and the language or languages licensed, are insinuated, not stated.Orwellians nowadays profess 'not to understand' their opponents and question their

'intelligibility'.21 Victims struggle vainly to express themselves in terms the Orwellian

will accept (indeed, some Orwellians can envisage no other kind of reply).22 This isintellectually objectionable, since despite frequent a-theoretic pretensions, the Orwellianrelies on some covert theory of meaning; a theory, moreover, which probably is false. Forthe very existence of an opposition to be dealt with indicates that the theory cannotaccommodate the linguistic data. Occasionally the tactic can lead to absurdities, whenOrwellian philosophers are torn between their polemical need not to understand theiropponent and their heartfelt conviction that they understand him all too well. Witnessthe blurb on Baker and Hacker's (1984) dig at Frege: 'By clarifying the mathematicalfoundations of Frege's logic, the authors render intelligible Frege's thinking what hedid. But by critical analysis they reveal much of what he thought to be unintelligible.'

Only in an Orwellian culture could such patent doublethink escape ridicule.23

And, of course, to any libertarian the tactic is morally objectionable, particularly in itssneaky post-war variant. Why should we put up with these linguistic tyrants? GSsuggest the right response: What are the meaning-rules to which we must conform?

14

What is this impoverished but ideologically pure language we must henceforwardspeak? And how are they both to be justified? This puts the onus back where it belongs -

on the aspiring linguistic legislator.24

'But surely some philosophers do talk nonsense, or fail to display that clarity which isthe good faith of a philosopher. How, then, are they to be dealt with if we are notallowed to not understand them? Can't they be brought to book?' Yes - but ourmethods must be piecemeal, not the wholesale slaughter dealt out by a partisansemantics. (For too many philosophers, Orwellian incomprehension is their only tool.)Piecemeal polemics are more effective anyway, since they convince others besides oursemantic partisans. Ad hominems, mockery, reductio ad absurdum – a whole battery ofnon-Orwellian techniques remain available. Where a thesis is vague or figurative we tryto make it clear and assess the truth of its clarified variants. If, after a sincere effort, wefind all the plausible interpretations false, then, and only then, are we entitled to saywith Quine that 'as to the thesis [in question] only further clarification can assure us that

this asserts anything at all' (1976, p. 106).25 For philosophic libertarians there is no shortway with bad philosophy.

Perhaps Quine's arguments can be revised so as to meet these constraints? The closedcircle of unintelligibility could be construed as an ad hominem. Philosophers tend toexpand the analytic by chalking up new 'conceptual truths'. These are often establishedthrough the 'unintelligibility' of their opposites. If the complex of ideas on which sucharguments depend is, by similar standards, 'unintelligible', this should be anembarrassment. But this argument is too sketchy to embarrass anyone. And though itmay be what Quine should have argued, it is a long way from what he actually said.

A more promising line is suggested by GS themselves. We might be harder on theseexpressions (i.e. the supposedly unclear members of the analytic circuit) if they wereeither technical terms or ordinary terms used in an extended sense (DD, p. 149). This isnot because of any vulgar Wittgensteinian aversion to technical terms in philosophy,but because the theory in which they are embedded is a failure. The idea behindanalytic truths is that they are true by something like fiat or linguistic convention. And

15

Quine has shown in 'Truth by Convention' that this analysis fails even for thoseparadigm analyticities, the truths of logic and arithmetic. It is rather as if chemists,post-Lavoisier, continued to use 'phlogiston' and a circle of interdefinable terms withoutgiving any account of how they intend to deal with Lavoisier's results. We are thusentitled to wonder whether what their philosophic counterparts say makes sense andeven to demand a little extra clarity.

The absence of an articulated and unrefuted theory of analyticity spells trouble for theother members of the circuit. 'Synonymy' and 'synonymous' are everyday words and,at the level of lexicography, are clear enough. No doubt 'bachelor' and 'unwed man'mean much the same. But if analyticities can be reduced to logic via the substitution ofsynonyms, there must be more to synonymy than this - at least if 'analytic' is to retain its'fifties range. To accept this account of analyticity, but stick with everyday synonymy,would leave us with a very restricted analytic - roughly, the Putnam-class plus logic.And this would undermine the GS-argument from consensus. So synonymy is used inan extended, theoretic sense. But no theory of extended or philosophic synonymy exists(once verification has been given up), and we have no pre-theoretic intuitions to go on.(Such intuitions as we have are the products of departed theories.) In thesecircumstances we may feel that synonymy 'stands in need of clarification'.

What about necessity, another member of the circuit (synonymy is defined assubstitutivity salva necessitate)? I don't think it can reasonably be claimed (at any ratenot nowadays) that the notion of necessity is unclear. Rather there are rival notions ofnecessity (codified by different logics) and rival accounts of these (a multiplicity of pureand depraved semantics). It is not intelligibility that is the problem, but a plethora ofinterpretations. Nevertheless, there is a problem about whether necessity can elucidatesynonymy. First because it is not clear that philosophic synonyms of the kind requiredcan be switched within the scope of a necessity operator without change of truth-value.(The lack of a theory of philosophic synonymy means we have no guarantee of this.)Second, expressions can be switched salva necessitate and not be (ordinarily)synonymous. Within some complete formulation of the propositional calculus, being atheorem (derivable from the axioms) and being a tautology (defined in terms of truth-tables) are necessarily coextensive. Yet the one is a proof-theoretic and the other a

16

semantic notion. They are not synonymous. This is clear from intentional contexts.Despite his understanding of both words, Greg may know that A is a tautology but notthat it is a theorem. (He is not aware of the completeness proof or is simply a duffer atlogic.) Nor can we define synonymy via substitutivity in intentional contexts. Humaninconsistency precludes such a project. At all events, necessity, though clear enough initself (or perhaps itselves) is no good for our present purposes.

Quine's original onslaught on 'analytic' and related words was a paradigm of Orwellianrhetoric. He demanded, in effect, that they be translated into a partisan Newspeak,based on absurdly strict semantic standards. Whether GS themselves employOrwellian tactics (and few philosophers in our time are entirely guiltless), they are rightto denounce them in others. However, Quine's criticisms can be revived without resortto an Orwellian theory of meaning. The words in question are theoretic; terms of thephilosophic art. They are therefore to be understood through the theories in which theyare embedded. These are non-existent. Clarification (in the form of a decent theory) isstill required.

VI. Breaking into the CircleGS also try to counter Quine's argument by breaking into the circle. They show howthey might introduce someone to the notion of logical impossibility by a sort ofinductive definition (a sequence of examples).

They could then move on to define its interdefinable relata. It is clearly conceptualimpossibilities, rather than bare and boring logical contradictions, that interest GS.Their example of a logical impossibility is this: 'My neighbour's three-year-old child isan adult' (DD, p. 150). There is, GS suggest, nothing coherent this could mean. Theybolster their case by excluding an obvious interpretation - that the child is a freak ofnature who has developed at enormous speed. (This is a rather cheap method ofmanufacturing conceptual impossibilities. You deny that a certain description can becoherently met and, when someone comes up with a plausible case, declare 'That is notwhat I meant!'. In effect, GS tacitly switch to another proposition which they likewiseassume to be impossible: 'My neighbor's three-year-old child, who is not a rapidly

17

maturing freak of nature, is an adult.') But there are other coherent interpretations of theoriginal statement besides this. The child could have been magically transformed intoan adult by a passing sorcerer. (Maybe there's nothing 'freakish' about this. Thesorcerer does it on such a grand scale that his victims [?] are granted adult rights on aroutine basis.) Or maybe the child is a legal adult since he understands Russell's Theoryof Types (unlikely, but logically possible, according to GS) and everyone who performsthis feat acquires adult status. Perhaps the child was born on Ganymede and is threeJovian years old. Or my neighbor is a bug-eyed monster (or 'bemmy' for short) whosechildren mature in three years in the natural course of events (non-freakishly, that is).Maybe the child is a time-traveler who has grown up during his journeyings but wasborn just three years ago. GS, it seems, confound inconceivability with their owninability to conceive.

GS might reply that some of the counterexamples to their alleged logical impossibilityare themselves impossible. Their surface plausibility masks a deep incoherence.Maybe, but this is no help to GS. For their aim was to elucidate 'logical' impossibility bydisplaying a paradigm case. If the paradigm seems to be possible after all - at least onsome interpretations - we cannot assume the concept is in order and use it to discreditthe counterexamples. Perhaps there are clear-cut cases of a conceptual impossibility onwhich an inductive definition could be based. But GS's example isn't one of them.

It is obvious that, for all their a-theoretic pretensions, GS remain reliant on a ghostlytheory. They nowhere dispute Quine's analysis of analyticity. Something is analytic orconceptually true if it can be reduced to a truth of logic through the substitution ofsynonyms. Thus something is logically impossible (it is analytic that not . . .) if it can bereduced to a logical contradiction through the substitution of synonyms. Do GS reallythink that this can be done with 'My neighbour's three-year-old child is an adult'?Presumably being more than three (earth?) years old is part of the meaning of 'adult'. But,as the above examples show, this is not a matter of uncontentious linguistic fact, a pieceof lexicographic trivia. The synonymy to which GS covertly appeal is thus synonymy inthe extended theoretic sense. Again we have a theoretic term without a stated theory.In its absence, it is not clear that they have marked out a cohesive class of impossibilitiesat all.

18

VII. Synonymy within the Bounds of Quineanism AloneGS show some sympathy for the latter part of Quine's paper. They are, perhaps,prepared to give up reductionist empiricism, the second of the two dogmas. Butsynonymy can be saved from the wreck. It can be defined in terms of Quine'sundogmatic empiricism. Two statements are synonymous 'if and only if anyexperiences which, on certain assumptions about the truth-values of other statements,confirm or disconfirm one of the pair, also, on the same assumptions, confirm ordisconfirm the other to the same degree' (DD, p. 156). But almost any pair of formallyconsistent statements can be converted into synonyms by this criterion. Take 'x iscordate' and 'x is renate' (to trot out Quine's two old warhorses). If we assume that allcordates are renates (and especially if we assume that this has maximum certainty),anything which confirms 'x is cordate' likewise confirms 'x is renate', and to the samedegree. But the co-extensiveness (even the necessary co-extensiveness) of the twopredicates does not convert the statements into synonyms, whatever GS may say.Perhaps their definition can be amended: 'Two statements are synonymous if on anyassumptions etc., experiences which would confirm or disconfirm the one, confirm ordisconfirm the other, and to the same degree.' On the first definition, synonyms wereembarrassingly abundant; on the second, they may be abolished altogether. Consider 'xis a creature with a heart' and 'x is cordate'. We can imagine the category cordateacquiring a special status in biology, becoming the name of a natural order. And wecan imagine cardiac surgery and prosthesis advancing to such a degree that a diseasedheart may be removed or replaced, not with a transplant or mechanical heart, but withseveral little pumps operating in different parts of the body. Mr X may be a true-borncordate, springing from hearty stock but, in his own person, entirely heartless. Evidencethat he is a cordate (genetic evidence for instance) will not confirm to the same degreethat he is a creature with a heart. Or consider 'x is a bachelor' and 'x is an unmarriedman'. We get in touch with bug-eyed monsters from the planet Skyro and discover that,apart from their anatomical oddities, they are much like ourselves; male and female,and given to legalized pair-bonding. Bachelor bemmies are surely possible; unmarried,male, but not men. Evidence could confirm X's bachelorhood without confirming hisunmarried hominity. A still more bizarre situation might allow unmarried men (at least

19

unmarried as men) who yet were not bachelors. Suppose that science establishes thatbemmies from Skyro are regularly reincarnated as human beings. The soul's path canbe traced in every case and memories of our earlier lives restored. Legal obligationscontracted as a bemmy might carry over into the new human life. Men, thoughunmarried, no longer count as bachelors if they married as bemmies (obligations tobemmy wives remain). Under these circumstances, confirming that X is an unmarriedman would not confirm to the same degree that X is a bachelor.

Don't these stories involve a meaning-change? Perhaps - though I am not at all surethat the sense of 'bachelor' shifts when applied to a bachelor bemmy. But GSs' projectwas to define synonymy (and hence meaning) with concepts acceptable to Quine. Theycannot, without circularity, import the notion of semantic change into Quine's meaning-theoretic vacuum. Neither their own definition, nor my amended version, will do.

VIII. Revisability and AnalyticityGSs' most startling claim (5) is that the analytic can be retained even if unrevisability isgiven up (DD, p. 157). It may well be that no statement is immune from revision in theface of experience. But the analytic and the synthetic can still be told apart, so long aswe distinguish the kind of giving up that consists in (merely?) admitting falsity fromchanging or dropping concepts. Nobody will deny (or if they do, nobody else willbelieve it) that the same words can be taken in different senses. When an analytic- or,more properly formerly analytic - statement is dropped, then 'the form of words inquestion changes from expressing an analytic statement to expressing a syntheticstatement' (DD, p. 157). It is the new synthetic variant that is discarded as false. Therecan be no change in the truth-value of our analyticity unless it ceases to be analytic -unless, that is, a meaning change transmutes it into a synthetic falsehood. The originalconcepts, which grounded the old analyticity, presumably remain available for use(though they will have to be clothed in different verbiage). That is, a new statementcould be concocted with the same sense and the same truth-value as the old, butexpressed in other words. Otherwise - if these concepts cannot be resurrected in theface of the facts - an analytic truth can be false or even nonsensical because the conceptsinvolved cannot be fitted into the correct world-view. The dogma defended by GS

20

would then be very different from the one in common currency. To maintain thedistinction in the face of universal revisability, it is necessary to maintain (a) that thereis a difference between rejecting a proposition as false and dropping the conceptsinvolved; (b) that whenever a supposed analyticity is rejected, this is because it haseither metamorphosed into a synthetic falsehood or because we have ceased to use theconcepts in question (it is not so much false as unspoken); and (c) that if the concepts arerevived (perhaps embodied in different words), the statement (or its equivalent) wouldagain be true. But this is to deny universal revisability in Quine's sense. True, on the (a)(b) (c) hypothesis there are no forms of words that cannot be dispensed with, and nostatements that cannot be left unsaid. But there are propositions which, given theirmeanings, cannot but be true when stated, even though nobody ever states them. Weare left with a banal and uncontroversial version of universal revisability. What Quinehad in mind was rather more radical.

The argument can be expressed as a dilemma. Either GS insist on (a) and (b) above, butnot (c), or they maintain (a), (b), and (c). If (a) and (b) are true but not (c), then ananalytic statement (and not just a sentence) can turn out not to be true, and not to be trueon its original reading. Its component concepts have no place in the new world-picture.Maybe the statement isn't strictly speaking false, but neither is it 'true in virtue of themeanings of words', nor unrevisable given those words (and that interpretation) - i.e. itis not analytic in the post-positivist sense. And this is to concede the case to moderateQuineans like Putnam. Suppose, on the other hand, that GS insist on (a), (b), and (c).Then the universal revisability they concede is quite innocuous, not at all the radicalthesis espoused by Quine. But the burden of their argument was that analyticity can be

defended in the face of Quine's subversive opinions.26

I can see two responses to this. The first is that it is not clear how Quine's universalrevisability differs from the innocuous kind. He has abandoned the concepts (ofsynonymy, etc.) that would enable him to make the needed distinctions. I do not thinkthis is correct (though it would visit Quine with a fitting nemesis). It seems obvious tome that there is a difference here, even though it is not obvious what that difference is.

21

A second, and more forceful, objection is that universal revisability in its strong form isfalse anyway. Putnam (1983, chs. 6-7) argues convincingly that there is at least onenecessary and a priori truth, namely, that not every proposition is both true and false.(He calls this the minimal principle of non-contradiction.) Putnam seems to think thatthere is an unrevisable core of mathematical and logical propositions which cannot berationally denied, and are thus far a priori. (What makes them true remains, however, amystery.) But this is no help to GS. For a good many of the 'analyticities' they woulddefend remain revisable in Putnam's eyes. Indeed, we can rework the above dilemmausing Putnam's limited revisability in place of Quine's universal variant. If GS insist on(a) and (b) alone, they grant Putnam everything he needs. If they hold out for (a), (b),and (c), then they are simply denying his limited revisability. Which means, in effect,that their analytic is incompatible with the cultural facts, and the History of Ideas.

IX. ConclusionThis essay, like that at which it is aimed, is mainly negative in purpose. GS defend adogma - that there is an analytic/synthetic distinction, and (more importantly) that itlies, more or less, where the positivists left it. This dogma is false; the phantom survivorof a dead system, with a baleful influence on the present. The practices to which GSappeal are superstitious, and their arguments are defective. Though a valid distinctioncan be drawn at least between analytic statements and others - what falls on the analyticside is uninteresting. There is no realm of deep but empty truths that are thephilosopher's special preserve. Like other scholars we are concerned with the world -though less directly than most.27

NOTES

1 All references to the reprint in Quine (1961).2 Putnam has written extensively on these subjects since 1962 (see Putnam [1983, chs. 5, 6, 7and 10]), but the changes he makes are not relevant to the present inquiry. In ch. 5 (pp. 196-7) hewrites as if logic can be included in the analytic, but he seems to be trying it on for size, not goingback on his earlier opinions ('Call these "analytic" if you like . . .'). He does recant his doctrine of(near) universal revisability in chs. 6-7, but this dogma does not directly concern us. Putnam usedto think - and still does in ch. 5 - that besides verbal banalities there are no absolutely a priori

22

statements. There are contextually a priori statements, statements which ground the theory ortheories we operate with, and which cannot be overthrown by observation alone. A new theoryhas to supersede the old one in which the earlier a priori 'truths' have no place. Of course, theremay be some statements which retain an a priori status in any true or plausibly true theory. As amatter of fact these are never challenged. They are still revisable for all that. In chs. 6-7, Putnamchanges his mind. There is at least one absolutely a priori truth; something it would never berational to deny - the minimal principle of contradiction: not every statement is both true andfalse. And maybe there are others. But though the minimal principle is true, what makes it trueremains a mystery. However, Putnam is still a friend to deviant logics. Just because some logicaltruths are absolutely a priori, this does not mean that all of them are.

Now the arguments of (1962) (at least as presented here) can withstand theseshifts. Though there may be immutable truths of logic, large chunks of it can still be chopped orchanged. It is still, in part, empirical, nor are the unrevisable bits true by convention (as the laterPutnam is concerned to stress). This is all that is required for his arguments to follow through.However, Putnam's revision of Quine is now more radical. He rejects the universal revisability ofnon-verbal statements (whereas formerly he was prepared to play along). This later view seemsto me correct. There are non-verbal statements we cannot sanely give up.3 For false or foolish analyticities, see his defense of Time Travel (Putnam [1979, ch. 15]) andhis famous attack on Malcolm in 'Dreaming and Depth Grammar' (1975, ch. 15).4 Nowadays he might be even more friendly to this suggestion, since he has developed somesympathy for relevance logic. See Putnam (1983, p. 131).5 Aquinas thought that if we understood God's nature we could see that he could not butexist. Unfortunately we don't have that kind of understanding. Anselm is wrong! (Of course,even if we did have such an understanding, it is not clear that the proof would be analytic in GSs'sense.) See Summa Theologica I. Q.2. a.l. (Aquinas [1945, pp. 20-23]).6 E.g. that it is 'evident' to our senses that some things in the world are 'in motion' (SummaTheologica I. Q.2. a.3, Aquinas [1945, p. 25]).7 See Summa Theologica I-II. Q.91, Aquinas (1945, pp. 616-27). Price's Review and Reid's Essayon the Active Powers of Man are excerpted in Raphael (ed.) (1969). Ross (1939) is by no means thelast believer in necessary but non-analytic moral truths, but he is perhaps the last mainstreamfigure to embrace this opinion. McCloskey's (1969) is the work of a philosophical heretic, as is theDivine positivism of Geach (1969).8 Schopenhauer (1966, 2, ch. IV).9 Frege (1953, pp. 101-2).10 See Schopenhauer (1965) where Kant's ethics is subjected to a savage critique.11 See Russell (1912, ch. 8). He says explicitly that not all a priori knowledge is analytic butleaves it rather vague how much of it is synthetic. It is interesting to see how Platonistic he is evenat this late date.12 Mill (1911, bk. II, chs. 4 and 5).

23

13 See Hume (1975, I. iv. 1, p. 20). Actually Hume vacillates about Geometry. In the Treatisesit does not reach the certainty of algebra and arithmetic. Hume (1978, I. iii. 1, pp. 71-72).14 Locke (1959, bk. IV, chs. VIII and II).15 See Russell (1937, p. v).16 This is very clear in Ayer (1971, ch. IV).17 In fact they overstepped the mark here. If 'explicating' verification-conditions were reallythe philosopher's special concern, no laboratory would be complete without one. Scientists wouldnot be able to work out what verifying experiments to perform. The positivists were ambitious tobe useful, but not as ambitious as this.18 See Ayer's comments in his autobiography (1977, p. 156), where he more or less accuses hisjuniors of living off verificationism's immoral earnings.19 To be fair, some philosophers were prepared to use Godel against logicism and expelarithmetic from the analytic. See Copi (1949).20 In Quine (1948) the ontological commitments of logicist theories - though not their variety -are stressed (Quine [1961, p. 14]). But Quine was aware of this too. The ontological peculiaritiesof Principia Mathematica as distinct from class-based systems are discussed in Quine (1961, pp.122-3), a paper of 1940s provenance.21 Davidson's description of his bland cultural imperialism (his refusal to understand thosewho disagree with him) as 'charity' is a brilliant Orwellian coup.22 See Dummett (1978, pp. 362-3). Dummett's realist about the past is a patsy who acceptswithout demur Dummett's theory of meaning and tries to show that what he wants to say can beexpressed in Dummettese- that conceptually impoverished fragment of English that could betaught in accordance with Dummett's ludicrously simple-minded theory of language-learning.23 The blurb-writer may feel he has protected himself against this charge by distinguishingbetween the intelligibility of Frege's thinking and the unintelligibility of his thought; between theactivity and the product. Thus it might be (humanly) understandable how someone came to be'speaking in tongues' at a revivalist meeting although the babble is incoherent. But this will notwash in Frege's case. For the doing in question, Frege's thinking, consists in concoctingpropositions and connecting them with others. If the actual propositions are unintelligible so isthe activity.24 I should stress that I am not impugning the honor of individual practitioners of theOrwellian art. It is the tradition that is corrupt, not they. The God of the philosophers willforgive them for they know not what they do.25 This is the concluding sentence of Quine (1936). Quine's efforts to elucidate the notion ofconventional truth were sincere. It was a doctrine to which he was much attracted and hadupheld just three months before. See Putnam (1983, p. 172).26 The argument in the last few paragraphs owes something to Dummett's discussion ofPutnam. Dummett (1978, p. 287).27 I am indebted to Alan Musgrave for his comments on an earlier draft.

24

REFERENCES

Aquinas, St. T. 1945. Introduction to St Thomas Aquinas, ed. by C. Pegis Anton. New York: ModernLibrary.Ayer, Sir A. J. 1971. Language, Truth and Logic, 2nd ed. Harmondsworth: Penguin.Ayer, Sir A. J. 1977. A Part of My Life. Oxford: Oxford University Press.Baker, G. P. and Hacker, P. M. S. 1948. Logical Excavations. Oxford: Blackwell.Copi, Irving M. 1949. 'Modern Logic and the Synthetic A Priori.' The Journal of Philosophy 46.Dummett, Michael 1969. 'The Reality of the Past.' Proceedings of the Aristotelian Society, n.s. LXIX.References to the reprint in Dummett (1978).Dummett, Michael 1978. Truth and Other Enigmas. London: Duckworth.Frege, G. 1953. The Foundations of Arithmetic, trans. J. L. Austin, 2nd ed. Oxford: Blackwell.Grice, H. P. and Strawson, P. F. 1976. 'In Defense of a Dogma.' Philosophical Review 65. .Hare, R. M. 1981. Moral Thinking. Oxford: Oxford University Press.Hume, David 1975. Enquiries concerning Human Understanding and concerning the Principles of Morals, ed. byL. A. Selby-Bigge/P. H. Nidditch, 2nd ed. Oxford: Oxford University Press.Hume, David 1978. A Treatise of Human Nature, ed. by L. A. Selby-Bigge/P. H. Nidditch, 3rd ed. Oxford:Oxford University Press.Locke, John 1959. An Essay concerning Human Understanding, ed. by Campbell Fraser, 2 vols. New York:Dover.McCloskey, H. J. 1969. Meta-ethics and Normative Ethics. The Hague: Martinus Nijhoff.Malcolm, Norman 1963. Knowledge and Certainty. Englewood Cliffs, N.J.: Prentice-Hall.Mill, J. S. 1911. A System of Logic. London: Longmans Green.Price, Richard 1974. A Review of the Principal Questions in Morals (1758), ed. by D. D. Raphael, 2nd ed.Oxford: Oxford University Press.Putnam, Hilary 1962. 'The Analytic and the Synthetic.' Minnesota Studies in Philosophy III. All references tothe reprint in Putnam (1975).Putnam, Hilary 1962. 'Dreaming and Depth Grammar', in R. Butler (ed.), Analytic and Philosophy: FirstSeries. Oxford: Blackwell. All references to the reprint in Putnam (1975).Putnam, Hilary 1968. 'The Logic of Quantum Mechanics' ('Is Logic Empirical?'), in R. Cohen and M.Wartofsky (eds.), Boston Studies in the Philosophy of Science 5, Dordrecht: Reidel. All references to thereprint in Putnam (1979).Putnam, Hilary 1975. Mind, Language and Reality. Philosophical Papers, 2. Cambridge: CambridgeUniversity Press.Putnam, Hilary 1979. Mathematics, Matter and Method. Philosophical Papers, 1, 2nd ed. Cambridge:Cambridge University Press.Putnam, Hilary 1983. Realism and Reason. Philosophical Papers, 3. Cambridge: Cambridge UniversityPress.

25

Quine, W. V. O. 1936. 'Truth by Convention', in O. H. Lee (ed.), Philosophical Essays for A. N. Whitehead.New York: Longmans Green. All references to the reprint in Quine (1976).Quine, W. V. O. 1948. 'On What There Is.' Review of Metaphysics. All references to the reprint in Quine(1961).Quine, W. V. O. 1951. 'Two Dogmas of Empiricism.' Philosophical Review 58. All references to the reprint inQuine (1961).Quine, W. V. O. 1954. 'Carnap and Logical Truth.' Synthese 12 (1960). All references to the reprint in Quine(1976).Quine, W. V. O. 1961. From a Logical Point of View, 2nd ed. New York: Harper & Row.Quine, W. V. O. 1976. The Ways of Paradox, 2nd ed. Cambridge, Mass.: Harvard University Press.Raphael, D. D. (ed.) 1969. The British Moralists. Oxford: Oxford University Press.Ross, Sir W. D. 1939. Foundations of Ethics. Oxford: Oxford University Press.Russell, Bertrand 1912. The Problems of Philosophy. London: Oxford University Press.Russell, Bertrand 1937. A Critical Exposition of the Philosophy of Leibnitz, 2nd ed. London: Allen & Unwin.Schopenhauer, Arthur 1965. On the Basis of Morality, trans. by E. F. J. Payne. Indianapolis: Bobbs-Merrill.Schopenhauer, Arthur 1966. The World as Will and Representation, trans. by E. F. J. Payne, 2 vols. NewYork: Dover.Wittgenstein, Ludwig 1967. Philosophical Investigations, 3rd ed., trans. by G. E. M. Anscombe. Oxford:Blackwell.

Received 5 September 1986

Charles R. Pigden, Department of Philosophy, University of Otago, P.O. Box 56,Dunedin, New Zealand.